You tested 1,000 different momentum signals on historical data. 15 of them showed a statistically significant positive return with a p-value under 0.01. You decide to allocate capital evenly across these 15 signals. Tell me why this will likely lose money, and how you would quantitatively adjust your inference process here.

你在历史数据上测试了 1000 个不同的动量信号，发现有 15 个信号的收益率达到了统计显著（p < 0.01）。你决定把资金平均分配给这 15 个信号。告诉我为什么这个组合大概率会亏钱？你会如何在定量层面上修正你的推断过程？